All cases of lung cancer diagnosed in the Tampere University Hospital catchment area in 1983-1987 were identified, analyzed for DNA flow cytometry and followed up to 1992. The patients were classified into 3 groups: screen-detected, symptom-detected, and detected by chance. The biological aggressiveness as indicated by DNA flow cytometry was not related to the survival of the symptom-detected patients. Also the screen-detected patients with an aggressive tumour (aneuploid or high S-phase fraction, SPF) had the same survival as the symptom-detected patients. The survival of screen-detected patients with a diploid or low SPF tumour was significantly better than that in the other groups. It is concluded that some of the previously known discrepancy of no effect on mortality and effect on survival of lung-cancer screening may be due to over-diagnosis, i.e., detection of morphologically malignant but biologically indolent lesions by screening.